(none)

[A.F.W.Coulson at edinburgh.ac.uk]

>
>How many fingers does a protein need to stick to DNA ?
>
>Whenever a molecular biologist gets a amino acid sequence, the first he will do
>is to look for signal peptides, nuclear transport signals, zinc fingers,
 leucine
>
>zippers etc.
>Now you can often read that a protein has one single zinc finger or a series of
>
>leucines fit "weakly" to a leucine zipper motif.
>I wonder whether these speculations have any significance.
>Now to my questions. When anybody detects such a motif, which tests or
>experiments have to be done to claim that a protein is really a leucine zipper
>or has zinc fingers.
>
This is a very important and topical point.  One can easily add many other
cases -- transmembrane helices, nuleotide binding domains, helix-turn-helix
DNA binding sites, etc  -- in which a similar process has gone on.  A
 structurall
/functional motif is identified in a small number of cases from X-ray structures
(where 'small number' is often 1 and sometimes 0) and then extended by sequence
analogy across a tottering structure of multiple sequence alignments until the
feature is being confidently 'identified' in every new sequence.  If a computer
program is used for this identification, then the only 'knowledge' which is
 brought
to bear is that which is encapsulated inthe computer program -- and this does
not yet extend to an assessment of structural, functional or biological
plausibility!

The type example of this process is protein secondary structure prediction.
Experts will tell you that the output of secondary struccture prediction
 programs
is only correct for about 60% of residues.  In other words, the predictions
are wrong almost as often as they are right.  Yet these results are often
presnted in sequencing papers as though they represented real new facts about
the protein.  I have even seen referee's reports on draft papers which insist
that secondary structure prediction should be carried out before the paper
is accepted!  It seems to be fatally easy to believe that the output of
a computer program conveys information about the real world.

There is a defence for the process, which goes like this:
'X-ray crystallography is a slow, laborious and expensive process which
may never be completed for many of the proteins of greatest interest
to molecular and cell biology.  Biolgists cannot afford to hang around
waiting for information which may never come.  You have to regard
"zinc fingers", "leucine zippers", etc, as so many suggestions which
are being kicked around to see if they are any use in interpreting
old experiments or designing new ones.  If they aren't any use, they'll
just be run over by next year's bandwagon.'

Take your pick which view you prefer; my experience is that people
become more sceptical about the significance of a motif, the more
their experimental work has a direct bearing on it........

Andrew Coulson